And this made me deeply confused. I work in marketing for a publishing company -- not one that has previously or is ever likely to publish McGuire, since we only do nonfiction -- and this struck me as precisely the opposite of how we typically operate. My employer is fairly laissez-faire with regard to street dates, particular when compared to the major fiction houses, but, still, my impression is that only a handful of books are published really day-and-date, and that the vast majority of books, even from major houses, have an official "on-sale" date, but are actually available to be sold as soon as cartons arrive at the bookseller, are opened, and the books are receipted into inventory.
(Real day-and-date publishing requires special shipments and printed cartons -- to make sure that stores know which books are to be held, and until when to hold them -- and my presumption is that Discount Armageddon shipped, with the rest of this month's DAW mass-market books, in unmarked cartons as usual. I could be wrong; perhaps someone will say so in comments.)
On top of that, when we have a book that we have strong expectations for, we regularly try to push for pre-orders (particularly on Amazon), since that leads to a stronger first week number and (usually) more copies continuing to go out for replenishment. So the idea of "don't buy my book now" sounded very backwards to me. (I also personally subscribe to the "get them to buy it now" theory of marketing in most situations -- if you put off the close until later, you lose a lot of your closes.)
I don't doubt that McGuire was told this by her publisher, and that she thinks this is vitally important. I'm not quite as sure that her publisher believes the same thing, though they might -- trade publishing is often run on expectations, consensus, and "we've always done it this way," so things that look counter-intuitive are often very common in the big houses. What I am pretty sure about is the actual direct effects of this early shipping. 
What a first-week sale can get you is one thing: onto a bestseller list, because that's likely to be the time of highest sales velocity for a mid-rank book. So I decided to look to see how likely that was. To call a mass-market book a "bestseller" these days really means one thing: that it charts on the New York Times mass-market list at least once. There's an additional wrinkle: the list as published in the Book Review lists only 20 books, but the "extended list" -- which is available online now, and was faxed to publishers back in the day before the Internet -- goes up to #35. And publishers are not shy about calling a book that hit #35 once a "bestseller," because it helps with sales, and we're in the business of selling books.
So I decided to compare McGuire's books -- both One Salt Sea, her most recent novel, and Rosemary and Rue, since that was her other series-starter for DAW -- with the books on the current New York Times list at #20 and #35. (Salt sold, in its peak week, 1.38 times Rosemary's peek week, so I'll use Salt as the comparison.)
The current week on the Times list isn't exactly the same as the most recent week captured by BookScan  -- and neither of those will be exactly the same as next week, when Discount Armageddon was expected to be published -- but it's all close enough for a rough sense.
So, the first thing I saw was that #20 (Wilbur Smith's Those in Peril) has remarkably high sales compared with books in adjoining positions (so high, in fact, that I think the Times is discounting or missing some sales, given that positioning). Given that, here's how some of the titles on the current NYT list compare to Salt's peak week:
- #1: Danielle Steel/44 Charles Street -- 9.59 Salts
- #15: George R.R. Martin/A Storm of Swords -- 1.56 Salts
- #20: Wilbur Smith/Those in Peril -- 3.18 Salts
- average of #19 (Nicholas Sparks/The Lucky One) and # 21 (Martin/A Feast for Crows) -- 1.52 Salts
- #35: Patricia Briggs/River Marked -- 0.95 Salts
So that's the real concern here: will this particular book hit a fairly arbitrary moving target, or fall short? (And it may be arbitrary, but it's a standard benchmark, so it is important.) Good luck to McGuire in hitting that level, whichever week it happens -- and, just maybe, Discount Armageddon will be popular enough to hit the list in more than one week!
 Indirect effects may, of course, include many other things, up to and including McGuire's worst fears -- if her publishing company really does look at first-week sales numbers as a priority (as opposed to, say, six-month sales numbers, or one-year sales numbers, or profitability at any of those points), it may have a deleterious effect. But I sincerely hope that isn't the case, because it would be stupid, and I still am enough of an optimist to think that publishing isn't actively stupid. Most of the time, anyway.
 BookScan captures, by general consensus, somewhere from 2/3 to 3/4 of the book outlets in the USA. Its numbers are proprietary, so I'll talk about them rather than give them directly -- and, again, they're generally lower than actual sales and are not the final numbers that publisher should be sharing with their authors.